Estimation of water stress of plant by deep learning of leaf images

2022 Fiscal Year Final Research Report

• Project/Area Number: 20K06328
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Review Section: Basic Section 41040:Agricultural environmental engineering and agricultural information engineering-related
• Research Institution: Toin University of Yokohama
• Principal Investigator: Sano Motoaki 桐蔭横浜大学, 医用工学部, 教授 (90206003)
• Co-Investigator(Kenkyū-buntansha): 杉本 恒美 桐蔭横浜大学, 工学研究科, 教授 (80257427)
• Project Period (FY): 2020-04-01 – 2023-03-31
• Keywords: 植物の水ストレス / 葉の固有振動数 / Webカメラ / 深層学習 / 最適灌水制御

Outline of Final Research Achievements

As a method for estimating water stress in plants, we have focused on the natural frequency of a leaf, and have shown that the diurnal behavior of the natural frequency of a leaf changes when the plant is subjected to water stress. We have also devised a rectangular region segmentation method for measuring the leaf frequency that does not require a region of interest for correlation tracking, and attempted to automate the measurement of natural frequency by using a convolutional neural network (CNN) to select a desired rectangular region.
As a result, we were able to determine the rectangular regions with almost 90% accuracy, and when we applied the CNN to draw a graph of the diurnal variation of the natural frequency of a leaf automatically, the results were almost the same as those produced by human judgment, although there were some outliers.

Free Research Field

信号処理

Academic Significance and Societal Importance of the Research Achievements

葉の固有振動数の日周変動がどのような機序で起きているのか、またその振る舞いは植物の種類によってどのように異なるのかなど、まだまだ調べなければならないことは数多く存在するが、何れにしても、植物の水ストレスが、葉の固有振動数の日周変化により推定できるということは学術的にも興味深い。
また、今回の研究をもとに、葉の固有振動数が自動的に計測できるようになり、それにより植物の水ストレスの状態が実時間かつ非侵襲的に推定できるようになれば、植物の最適灌水制御システムに向けての大きな前進となり、土耕栽培の植物工場などへの応用が期待される。